Senior AI Scheduler Engineer: Kubernetes & GPUs

Firmus; Singapore; On-site; SGD 180,000 - 240,000; Full time

Job summary

Firmus Technologies in Singapore is seeking a Senior AI/Platform Engineer to lead the Kubernetes-native scheduler development for large-scale AI factories. You will design, implement, and operate the Model-to-Grid workload management stack, enabling topology-aware placement, fault-tolerant execution, and high GPU utilization.

Collaborate with AI, inference, platform, and security teams to deliver scalable, multi-tenant scheduling with strong observability and CI/CD pipelines, contributing to a

Qualifications

• 5+ years in DevOps, platform engineering, distributed systems, cloud infrastructure, or similar roles.
• Deep Kubernetes expertise including control planes, scheduling, controllers, and multi-tenancy.
• Experience building or extending workload schedulers or orchestration systems (Kubernetes Scheduler Framework, etc.).
• Strong Go programming with Python for automation and tooling.
• Experience with GPU workloads, scheduling, and topology-aware placement.
• Proficiency with observability, performance analysis, and SLOs, using Prometheus, Grafana, OpenTelemetry, and related tools.
• CI/CD, GitOps, IaC and safe production deployments (GitHub Actions, GitLab CI, Argo CD, Terraform, Helm).
• Security-conscious multi-tenant platform engineering and incident response.

Responsibilities

• Design, build, operate and improve the Kubernetes-native scheduling platform for AI workloads.
• Develop CRDs, controllers, admission webhooks, plugins, APIs and automation for workload submission and management.
• Define scheduling policies for training, fine-tuning, inference, benchmarking, and batch workloads.
• Implement topology-aware placement considering GPU locality, NVLink/NVSwitch topology, and storage locality.
• Develop AI-factory resource-aware scheduling aligned with capacity, health, power, thermal, and maintenance signals.
• Integrate scheduler with Kubernetes, GPU plugins, network and storage services, and observability systems.
• Build scheduler templates, model recipes, and self-service workflows for workload submission and troubleshooting.
• Establish CI/CD, GitOps, and operator experiences for scheduler components.

Skills

Kubernetes

Go

Python

CI/CD

GitOps

Distributed systems

GPU workloads

Topology-aware scheduling

Observability

Multi-tenant

Tools

Slurm

Slinky

Kueue

KAI

Volcano

YuniKorn

Run:AI

GitHub Actions

GitLab CI

Argo CD

Flux

Terraform

Helm

Kustomize

OPA

Kyverno

Job description

Firmus Technologies in Singapore is seeking a Senior AI/Platform Engineer to lead the Kubernetes-native scheduler development for large-scale AI factories. You will design, implement, and operate the Model-to-Grid workload management stack, enabling topology-aware placement, fault-tolerant execution, and high GPU utilization.

Collaborate with AI, inference, platform, and security teams to deliver scalable, multi-tenant scheduling with strong observability and CI/CD pipelines, contributing to a